This simplifies the API and hides an unnecessary implementation detail.

Add `error` finish reason

Support `reasoning` in ChatStreamResult deltas

Add parsing options for valueNotFound case

Add gpt_4o_transcribe and gpt_4o_mini_transcribe models.

Add init for audio options.

Fix call order and write more tests

Audio stream fix.

Tests on Linux complained

Seems to be required on Linux

Fix "expiresAt" parameter of audio delta.
Fix StreaminSession onComplete calling - onComplete called before all chunks produced.

Update README.md

Feat: Get token usage data for streamed chat completion response.

# Conflicts:
#	Sources/OpenAI/Public/Models/ChatQuery.swift
#	Sources/OpenAI/Public/Models/ChatStreamResult.swift

Make `created: TimeInterval` optional

Add ParsingOptions and use for ChatStreamResult

Added support for audio-preview models, including updates to query pa…

To fix parsing Gemini responses

Added support for audio-preview models, including updates to query parameters, message parameters, and audio delta.